Work has started on a Pharmaceutical Oncology Initiative (POI) study looking at statistical methods available for adjusting time-to-event estimates in the presence of treatment crossover. Treatment crossover can cause bias in the intention to treat estimation of treatment effects, typically because patients in the control group switch onto the new treatment and benefit from it. Statistical methods such as structural nested failure time models and inverse probability of censoring weights will be tested through simulation studies and real world data applications, with the objective of identifying the most suitable method to use in a range of different situations.
The study will extend work recently published by the study team: Morden JP, Lambert PC, Latimer N, Wailoo A, Abrams K. Assessing methods for dealing with treatment switching in randomised controlled trials: a simulation study, BMC Med Res Methodol 2011;11:4.
Project team is: Nick Latimer, Allan Wailoo, Keith Abrams, Paul Lambert, James Morden and Patrick Fitzgerald. Contact Nick for further details.